Skip to content

Latest commit

 

History

History
140 lines (108 loc) · 6.43 KB

readme.md

File metadata and controls

140 lines (108 loc) · 6.43 KB

Dremio AI Chat

Getting Started

  • fork/clone the repo
  • cd into the repo
  • create a virtual environment python -m venv venv
  • activate the virtual environment source venv/bin/activate
  • install the requirements pip install -r requirements.txt
  • run the app python app3.py
  • open your browser and go to http://localhost:5000

Comparison of app.py/index.html, app2.py/index2.html, and app3.py/index3.html

Overview

The Dremio AI Chat is a Flask-based web application that integrates OpenAI's GPT-4 with real-time data retrieval from Dremio. It allows users to query sales opportunities dynamically. The three versions of the application—app.py/index.html, app2.py/index2.html, and app3.py/index3.html—differ in how they handle data retrieval, user interaction, and session management.

Comparison of app.py, app2.py, and app3.py

1. User Input and Context Retrieval

  • app.py

    • Users submit a question only.
    • On the first request, a predefined SQL query fetches general data from Dremio.
    • The retrieved data is injected once into the AI prompt.
    • All subsequent responses reuse the same dataset.

  • app2.py

    • Users submit a customer name and a question.
    • The app first queries Dremio to find the customer’s unique ID.
    • Then, it uses that ID to retrieve customer-specific data.
    • The AI assistant uses this data to generate a personalized response.

  • app3.py

    • Users submit a single natural language question without specifying a customer.
    • The AI agent decides when to:
      1. Retrieve a customer list (to verify customer name spelling).
      2. Find the customer ID.
      3. Fetch customer-specific data from Dremio.
    • The AI dynamically calls the necessary tools to retrieve and incorporate data into its response.
    • Context from previous messages is preserved, allowing for a continuous, conversational experience.

2. Querying and Data Integration

  • app.py

    • Executes one SQL query on the first request.
    • Uses a preloaded dataset for all responses.

  • app2.py

    • Uses a two-step SQL query process:
      1. Retrieve customer ID.
      2. Fetch data for that specific customer.
    • Ensures responses are tailored to a specific customer.

  • app3.py

    • Uses LangChain tools to execute on-demand queries.
    • The AI agent determines when to query for customer names, IDs, or customer-specific data.
    • Responses dynamically adapt based on retrieved data.

3. Chat Session and Context Management

  • app.py

    • Maintains chat history for a session.
    • Resets on page refresh, requiring the first query to run again.

  • app2.py

    • Maintains chat history during a session.
    • Allows users to start a new conversation by entering a new customer name.
    • Resets on refresh.

  • app3.py

    • Maintains a continuous chat experience, preserving context across responses.
    • Chat history accumulates naturally like a real chat app.
    • Resets on refresh, but within a session, it remembers previous exchanges.

4. Frontend and User Experience

index.html (Used with app.py)

  • Simple chat UI with a single input field.
  • The user can only ask general questions based on the preloaded dataset.
  • Uses a clean aquatic UI with a chat history display.

index2.html (Used with app2.py)

  • Includes two input fields: one for customer name and one for the question.
  • Introduces a "Typing..." loading animation.
  • AI responses are formatted with <pre> tags to ensure structured Dremio data is readable.

index3.html (Used with app3.py)

  • Chat input remains at the bottom (like modern messaging apps).
  • Messages accumulate correctly instead of replacing previous ones.
  • AI automatically determines whether it needs to fetch additional data.
  • Auto-scrolls to the latest message for a seamless experience.

📊 Summary of Key Differences

Feature app.py / index.html app2.py / index2.html app3.py / index3.html
User Input Single question field Customer + question fields Single natural language question
Data Querying One-time query Two-step query (customer lookup + data retrieval) AI calls tools to fetch customer details as needed
Response Context Static dataset Customer-specific data per request Dynamic, AI-driven retrieval of necessary data
Personalization Generic responses Personalized responses per customer Fully adaptive to user needs
Session Handling Resets on refresh Resets on refresh, allows customer changes Preserves context within session, resets on refresh
Frontend Features Simple chat UI Customer field, loader animation, formatted response Full chat-like experience, messages accumulate, auto-scroll

🚀 Choosing the Right Version

✅ Use app.py/index.html if:

  • Responses should be based on a static dataset.
  • Users will ask general questions.
  • Performance is a priority (no need for multiple queries).

✅ Use app2.py/index2.html if:

  • Responses should be customer-specific.
  • Users should manually specify the customer name.
  • The assistant should fetch relevant data dynamically before answering.

✅ Use app3.py/index3.html if:

  • You want a fully conversational chat experience.
  • Users should not need to specify customer details manually.
  • The AI should determine when and how to fetch data dynamically.
  • Context from past messages should be remembered for a continuous conversation.

🎯 Final Thoughts

Each version provides a real-time AI-powered chat experience backed by Dremio's data engine.

  • app.py is the simplest, but lacks personalization.
  • app2.py improves personalization, but requires manual input for customer selection.
  • app3.py is the most advanced, allowing fully dynamic and context-aware conversations.

🚀 Choose the right version based on your needs! 🚀